ABBREVIATiONS: AR, adrenergic receptor(s); ISA, intrinsic sympathomimetic activity; CYP, cyanopindolol; CHW, Chinese hamster fibroblast(s).
490
0026.895X/94/030490. 10$03.00/0
Copyright © by The American Society for Pharmacology and Experimental Therapeutics
All rights of reproduction in any form reserved.MOLECULAR PHARMACOLOGY, 45:490-499
Inverse Agonist Activity of �3-Adrenergic Antagonists
PETER CHIDIAC, TERENCE E. HEBERT, MANON VALIQUETTE, MICHAEL DENNIS, and MICHEL BOUVIER
Departement de biochimie and Groupe de Recherche sur !e Systeme Nerieux Autonome (P.C., T.E.H., M. V., MB.) and D#{233}partementdePharmacologie (M.D.), Universit#{233}de Montr#{233}al,Montr#{233}al,H3C 3J7 Canada, and Biosignal Inc., Montr#{233}al,H3J 1R4 Canada (M.D.)
Received August 23, 1993; Accepted December 7, 1993
SUMMARYAgonist-independent properties of the human f32-adrenergicreceptor (fl2AR) were studied using the baculovirus expressionsystem in Sf9 cells. In the absence of agonist but in the presenceof GTP, membranes from cells expressing the fl2AR exhibitedhigher levels of cAMP production than did membranes fromuninfected cells or from cells infected with wild-type baculovirus.The increase in cAMP production was proportional to the numberof /32AR expressed, up to 40 pmol/mg of membrane protein, andit could be inhibited in a dose-dependent manner by I3AR antag-onists. The increase and its reversal both were independent ofthe possible presence of contaminating catecholamines in theculture medium and thus appear to reflect spontaneous fl2ARactivity and direct antagonist-receptor interactions, respectively.The maximal level of inhibition varied among the flAR ligandstested, to yield the following rank order of “inverse efficacy”:timolol � propranolol > alprenolol � pindolol > labetalol >
dichloroisoproterenol. The same rank order was observed usingmembranes prepared from Chinese hamster fibroblasts express-ing /32AR. The effect of timolol was partly blocked by labetaloland dichloroisoproterenol, in an apparently competitive manner.The intracellular cAMP content of Sf9 cells cultured in serum-free medium was also increased by the expression of fl2AR, andthat increase was reversed by timolol and propranolol, consistentwith observations in membrane preparations. The propertiesrevealed by the expression of the fl2AR in Sf9 cells suggest twoagonist-independent traits of G protein-linked receptors, i.e., 1)that unliganded receptors are able to activate G proteins both inmembrane preparations and in whole cells and 2) that antago-nists may mediate their effects not only by preventing the bindingof agonists but also by decreasing the propensity of the receptorto assume an active state.
Antagonists inhibit the binding of agonists to receptors, and
the physiological effects of the former usually are attributed to
their ability to prevent activation of receptors by endogenoushormones and neurotransmitters. In general, it is thought that
antagonists do not modulate the activity of unliganded recep-
tors, but a growing body of evidence contradicts this notion.
Antagonists have been reported to produce effects opposite to
those of the corresponding agonists at both benzodiazepine
receptors and G protein-linked receptors (reviewed in Refs. 1
and 2, respectively). The direct, agonist-independent modula-tion of receptor activity by antagonists has been referred to
variously as inverse agonism (3), negative or reverse intrinsic
activity (2, 4), and negative antagonism (5).
At the benzodiazepine receptor/�y-aminobutyric acid type A
receptor complex, benzodiazepine agonists allosterically in-
crease the affinity of “y-aminobutyric acid, whereas inverse
agonists have the opposite effect (3). The biochemical proper-
ties of benzodiazepine agonists are reflected in vivo by anxiol-
ytic and anticonvulsant effects. Inverse agonists, in contrast,
This work was made possible by grants from the Heart and Stroke Foundationof Canada and the Medical Research Council of Canada. P.C. and T.E.H. holdfellowships from the Heart and Stroke Foundation of Canada. MB. is a scholarof the Medical Research Council of Canada.
tend to be anxiogenic or proconvulsant (1, 6). It has been
proposed that the physiological effects of agonists and inverse
agonists in this system are correlated with their ability to
stabilize either the “active” or the “inactive” form of the recep-
tor complex, respectively (3, 6).
Similarly, antagonists have been reported to produce effects
opposite to those of agonists at G protein-linked receptors, withrespect to both second messenger (7) and G protein regulation
(4, 8, 9). However, these effects are not well characterized andhave been reported in reconstituted or broken cell preparations
but not in intact systems. Thus, in contrast to the benzodiaze-
pine receptor, a physiological correlate to the biochemical ef-fects of inverse agonists appears to be lacking, and the relevance
of agonist-independent effects of antagonists at G protein-
coupled receptors is unclear.
Presumably, inverse agonism requires that G protein-linked
receptors can exhibit spontaneous, agonist-independent activ-
ity, and some evidence suggests that this can occur in intact
systems (10). It has been suggested that the demonstration of
inverse agonism at G protein-linked receptors in such systems
would lend support to the hypothesis that hormone-independ-
ent effects contribute to the biological responses observed with
antagonists (2).
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
Inverse Agonism at fi2AR 491
In the present study we show that the human f32AR increases
the production of cAMP both in Sf9 cells and in membranes
derived from those cells, in an agonist-independent manner,
and that this increase can be inhibited by $AR antagonists. We
also show that similar effects occur in membranes derived from
mammalian cells expressing fl2AR. These observations suggest
that the mechanism of action of $AR antagonists in vivo may
be, in addition to preventing hormone binding, to decrease the
propensity of the receptor to assume a conformation that is
required for the stimulation of cAMP production.
Materials and Methods
Experimental Procedures
Cell culture and infection with recombinant baculovirus. Sf9cells were cultured at 27’ in Grace’s supplemented insect medium
(GIBCO), to which were added 10% fetal bovine serum (PA. Biolog-
icals, Sydney, Australia), 0.1% Pluronic F-68 (GIBCO), 50 �g/mlgentamicin sulfate, and 2.5 �zg/ml fungizone. For some experiments,
serum was omitted from the culture medium for the final 18 hr of
infection. For most experiments, cells were grown in 100-ml spinner
flasks; in some cases, cells were grown as monolayers in 75-cm2 flasks.Cells were infected with wild-type or recombinant baculovirus when
they reached a density of 1.0-4.0 x 106 cells/ml and were harvested 48
hr after infection, unless indicated otherwise. Experiments were carried
out either on preparations of washed membranes or on whole cells.
Preparation of washed membranes from Sf9 cells. Washedmembranes were prepared essentially as described previously (11).Infected cells (10 or 20 ml) were washed twice with 2 volumes of ice-
cold phosphate-buffered saline and lysed with a Polytron homogenizer
(two 5-sec bursts) in 10 ml of cold buffer containing 5 mM Tris . HC1,pH 7.4, 2 mM EDTA, 10 �g/ml benzamidine, 5 �g/ml soybean trypsin
inhibitor, and 5 �tg/ml leupeptin. Lysates were centrifuged at 500 x gfor 5 mm at 4” , and the pellet was resuspended in 4 ml of buffer,
homogenized, and centrifuged again. The two supernatants were pooledand centrifuged at 45,000 x g for 20 mm at 4’, and the pellet was
washed twice in 10 ml of the same buffer. The membranes were
resuspended in buffer containing 75 mM Tris . HC1, pH 7.4, 12.5 mM
MgCl2, 2 mM EDTA, 10 �g/ml benzamidine, 5 �tg/ml soybean trypsin
inhibitor, and 5 j�g/ml leupeptin and were then used for adenylyl cyclaseor binding assays. Protein content was determined according to the
method of Bradford (12).Culture of mammalian cells and preparation of membranes.
CHW cells expressing fl2AR (4-5 pmol of receptor/mg of protein) were
cultured and membranes were prepared as described previously (13).Assay of adenylyl cyclase activity. In membranes, assays were
carried out using 5-10 �ig of protein with 0.12 mM ATP, 1-2 x 106
cpm/assay tube of [a-32P]ATP, 0.10 mM cAMP, 53 �M GTP, 2.7 mMphospho(enol)pyruvate, 1.0 IU ofmyokinase, 0.2 IU ofpyruvate kinase,
4 �g/ml benzamidine, 2 �sg/ml soybean trypsin inhibitor, 2 �g/ml
leupeptin, and varying concentrations of AR ligands, in a total volumeof 50 �l. Samples were incubated at 37” for 15 or 30 mm, and thereaction was terminated by addition of 1 ml of a cold solution contain-ing 0.3 mM cAMP, 20,000 cpm of [3HJcAMP, and 0.4 mM ATP. [32P1
cAMP was separated by chromatography using a Dowex gel, followedby aluminium oxide (14).
Assay of intracellular cAMP. Intracellular cAMP was determined
by competition with [3HIcAMP for a specific binding protein (15). Cellsthat had been infected for 48 hr with either wild-type or recombinant
baculovirus encoding the human fi2AR were deprived of serum for thefinal 18 hr of infection and then used for experiments. One-milliliteraliquots containing 4 x 106 to 1 x iO� cells were treated for 30 mm at
ambient temperature with various ligands in serum-free medium.cAMP was recovered from the cells as described previously (16) (Amer-sham kit no. TRK.432). One sample was prepared for each experimen-
tal condition, and each was assayed in duplicate for cAMP. Cellular
cAMP was estimated by combining 50 Ml of supernatant with L3HIcAMP (4.5 nM final concentration) and allowing the labeled and
unlabeled forms of the nucleotide to compete for a high affinity cAMP-
binding protein for 2 hr at 4”; unbound [3H]cAMP was removed by
adsorption to charcoal followed by centrifugation, and the amount of
radioactivity from the supernatant was determined. The amount of
cAMP in each aliquot was estimated from a standard curve determined
with samples that ranged from 1 to 16 pmol of cAMP/assay tube.
Binding of AR uganda. Washed membranes were labeled with‘25ICYP to study the binding properties of the �92AR. For determina-tions of capacity, membranes were incubated with a saturating concen-tration of the radioligand (�300 pM; K,, = 33 pM) in the absence orpresence of iO� M alprenolol, with the difference in bound ‘�I-CYP
between the two conditions being taken as the total specific binding.
The inhibition of binding of a subsaturating concentration of ‘251-CYP
(169 ± 8 pM) by increasing conceiitrations of unlabeled AR ligands was
used to estimate the binding affinities of the latter compounds. In each
case, assays were carried out in triplicate and membranes were incu-
bated for 1.5 hr at ambient temperature in buffer containing 75 m�i
Tris, pH 7.4, 12.5 mM MgCl2, 5 m�i EDTA, 10 ��g/ml benzamidine, 5�zg/ml soybean trypsin inhibitor, and 5 �sg/ml leupeptin. Unbound
radioligand was removed by rapid filtration through Whatman GF/C
filters, which were then rinsed three times with 2 ml of 25 mM Tris,
pH 7.4, at 4”.
Analysis of Results
The concentration dependence of adenylyl cyclase activity on various
AR ligands in membranes was analyzed according to a four-parameterlogistic equation analagous to the Hill equation (ALLFIT) (17). For all
analyses, the slope factor was fixed at 1. For each flAR ligand listed inTable 1, replicate experiments were fitted simultaneously with EC�
common to all sets of data, allowing the asymptotes (i.e., y� and Yr-.
�) to vary for individual sets of data. The fitted values of Yx-o and �
were used to define 0s and aj, respectively, for the calculation of �
(see below). The inhibition of ‘251-CYP binding by IIAR antagonists
was analyzed according to the multisite model by nonlinear least-squares regression, as described (18). Data were analyzed assuming a
single class of binding sites. For each unlabeled ligand, data from all
experiments were fitted simultaneously with the affinity (Kd) common
to all sets of data. Estimates of nonspecific binding and total receptor
concentration were allowed to vary among individual sets of data.
Unless specified otherwise, values listed in the text, table, and figurelegends represent averages ± standard errors, and such values were
compared where appropriate using Student’s t test; for fitted values of
Kd, EC�, y�, and Yx-o, the accompanying error reflects the range overwhich the sum of squares of the residuals is insensitive to the value of
the parameter.
Inverse efficacy (E5�) is defined as the fraction of spontaneous (i.e,
agonist-independent) receptor activity that can be inhibited by a givenagent, as shown in the equation E1�� = Al/AR, where A, = (a� - aj)/as
and AR (as - a�)/as. AR represents the relative signal that is attrib-utable to the receptor alone, i.e., the difference between the signal
observed in the presence of receptor but in the absence of agonist or
antagonist (as) and that observed in the absence of receptor (an),divided by a5. A, represents the relative decrease induced by maximal
concentrations of a given agent, i.e., the difference between a� and thesignal observed at a saturating concentration of ligand (a,), divided
by as.
In the present study, a5 and a, represent cAMP production in theabsence of AR ligand and in the presence of a maximally effective
concentration of the latter, respectively. These were estimated fromsimultaneous analyses, as described above, for experiments carried outin membranes; for experiments carried out in whole cells the measured
values were used. The parameter so represents the rate of cAMP
production observed with preparations lacking fl2AR in the absence of
any AR ligand. The mean values ofAR were 0.78 ± 0.02 in membranes
(11 experiments) and 0.64 ± 0.17 in whole cells (four experiments), and
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
C
EC0
0
Ea.2
U
0E0.
0 10 20 30 40Receptor level (pmol/mg)
50
Effect of human I9AR expression on the production of
cAMP. In washed membranes prepared from Sf� cells, the
level of cAMP produced in the presence of 53 �M GTP was
found to be greater with cells infected for 48 hr with baculovirusencoding the human fl2AR than with either cells infected with
wild-type baculovirus or uninfected cells (Fig. 1). To establish
that the increase in cAMP was not due to stimulation of thereceptor by contaminating catecholamines that may have beenpresent in the complete medium, experiments were carried out
on cells maintained in serum-free medium for the final 18 hr
and treated with alprenolol (1 �tM) for the final 30 mm of the
infection. As shown in Fig. 1, inset, the receptor-mediated
increase was still observed under these conditions. An increase
was also observed with serum-deprived cells that had not been
treated with alprenolol (data not shown). It follows that the
f32AR is spontaneously active in washed membranes. In addition
to the receptor-mediated, agonist-independent increase in ad-
enylyl cyclase activity observed in membranes from cells ex-
pressing �92AR, the flAR agonist isoproterenol further stimu-
lated the enzyme. The agonist was without effect on membranes
from cells that did not express the receptor (Fig. 1).
Infections for 48 hr at various multiplicities of infection
yielded levels of fl2AR expression ranging from 0.05 to 45 pmol/
mg of protein. As shown in Fig. 2, the agonist-independent and
the net isoproterenol-stimulated and forskolin-stimulated in-
creases in cAMP production all were linearly correlated with
the level of $2AR expression (r� = 0.97, r� = 0.99, and r� = 0.98,
respectively).
160
� 140
E
I 120
2 1000.
E 80
a-� 60
C)
-o �#{176}EQ- 20
0
30
150 -
125 -
100 -
75 -
50
25
0
>�
>
C)
a)a,0 �-.
0
at,Q-a)
0
.� E
20
10
I
n0
uninfected fl�AR
uninfected wt virus �2AR
Inhibition of receptor-stimulated cAMP production.
Alprenolol partly reversed the increase in cAMP production
associated with the expression of the fl2AR, whereas isoproter-
enol stimulated adenylyl cyclase activity >2-fold in the same
preparations (Fig. 3). When membranes containing fl2AR were
treated with both isoproterenol and alprenolol, a net decrease
was observed in the level of enzymic activity (data not shown),
similar to that found with alprenolol alone. Membranes from
cells lacking the receptor exhibited similar levels of enzymic
activity in the presence and absence of alprenolol (p = 0.82),
indicating that the inhibition was due to an interaction between
492 Chidiac at a!.
these were taken as constants in further calculations. For both mem-
brane and whole-cell assays, values of A1 for individual ligands were
averaged, and each value of � was calculated by dividing the mean
value of A, by the mean value of AR.
Results
Fig. 1. Activity of adenylyl cyclase in membranes from uninfected Sf9cells and Sf9 cells infected for 48 hr with either wild-type (wt) baculovirusor recombinant baculovirus encoding the human fl2AR. The productionof CAMP was measured in the absence (0) or presence of 10 �
isoproterenol (U). The data shown represent means from at least threeexperiments carried out in duplicate. Inset, the production of cAMP wasmeasured in membranes from cells that had been deprived of serum forthe final 18 hr of infection and treated with 1 �M aiprenolol for 30 mmbefore harvesting.
Fig. 2. Effect of receptor level on GTP-, isoproterenol-, and forskolin-stimulated adenylyl cyclase activity. The level of fl2AR in membraneswas varied by infecting Sf9 cells for 48 hr at increasing multiplicities ofinfection, and receptor levels were determined as described in Materialsand Methods. Production of cAMP was measured in the presence of 53�tM GTP alone 4 or with the additional inclusion of 10 �M isoproterenolor 100 zM forskolin. The net stimulation by isoproterenol (0) and forskolin(#{149})was determined by subtracting the level of activity observed in thepresence of GTP only, at each concentration of receptor.
Isoproterenol +
Aiprenolol +
Fig. 3. Effect of alprenolol on fl2AR-stimulated adenylyl cyclase activity.The production of cAMP in membranes from Sf9 cells lacking fl2AR (i.e.,either uninfected or infected with wild-type baculovirus) was subtractedfrom that in membranes expressing the receptor, in the presence ofisoproterenol (1 0 �zM), alprenolol (1 0 SM), or neither. The data shownrepresent means from four experiments.
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
100
80 �
0
60
40
20
0
aiprenolol and the receptor. To ensure that competition with
serum catecholamines in the culture medium was not respon-
sible for the observed inhibition, experiments were carried outon cells maintained in serum-free medium. Alprenolol was
found to inhibit adenylyl cyclase in a dose-dependent manner
in washed membranes from cells cultured under such conditions
(Fig. 4). This inhibition was observed even when cells weretreated with aiprenolol for 30 mm before harvesting. Similarly,
treatment of cells maintained in complete medium with 1 �iM
alprenolol for 30 mm before harvesting of membranes did not
prevent the decrease in cAMP production with alprenolol (data
not shown), indicating that the inhibition observed under nor-
mal experimental conditions was not dependent upon contam-inants in the medium.
Propranolol also was found to inhibit the fl2AR-associatedincrease in adenylyl cyclase activity (Fig. 5). The magnitude of
the response appeared to be limited by the receptor-inducedincrease in cAMP. In absolute terms, the inhibition of the
enzyme by propranolol (Fig. 5) or alprenolol (data not shown)
increased with the receptor-dependent increase in cAMP. Atthe lowest concentrations of receptor investigated ( 12-hr infec-tion, 0.03 ± 0.02 pmol of fl2AR/mg of protein), the inhibition
could not be observed reliably, although a small amount ofstimulation was observed under such conditions in response to
isoproterenol (data not shown). The maximal levels of both thestimulation and the inhibition of adenylyl cyclase thus appearto be dependent upon the density of fl2AR in the membrane.
A variety of AR ligands were characterized with respect to
agonist-independent inhibitory potency, binding affinity, and
inverse efficacy (Fig. 6; Table 1). The production of cAMP in
>“
>
C)
0
a,0C)>‘C)
>‘Ca)�00
a).0
0
.0
C
-6 -5
I � I I � � I
I I I � -U
100
75
50
25 -
0-
40
30
20
10
0
0
-11 -10 -9 -8 -7 -6 -5
-10 -9 -8 -7 -6 -5
Inverse Agonism at �92AR 493
log [aiprenolol]
Fig. 4. Inhibition of agonist-independent fl2AR-stimulated adenylyl cy-clase by alprenolol in membranes from cells deprived of serum. Sf9 cellsinfected with baculovirus and cultured in serum-free medium for the final18 hr of infection were treated with vehicle (0) or 1 � alprenolol ((s) for30 mm before membrane preparation. The production of cAMP wasmeasured at the concentrations of aiprenolol shown on the abscissa.The data shown are from a representative experiment (two determina-tions) and were normalized to the fitted asymptotes, which were asfollows: estimated cAMP production in the absence of &pren�ol (y�)was 45 ± 1 and 75 ± 1 pmol/mg/min in alprenolol-treated and controlmembranes, respectively, and that in the presence of a maximally inhib-iting amount of alprenolol (Y�.��) was 30 ± 1 and 34 ± 1 pmol/mg/min inalprenolol-treated and control membranes, respectively.
a:
Ea:a)0
0�
E
a-
C)
0
Eci.
-11 -10 -9 -8 -7
log [propranolol]
Fig. 5. Effect of time of infection on the inhibition of membrane adenylylcyclase activity by propranolol. Sf9 cells were infected with baculovirusencoding the fl2AR for 0 (a), 12 (0), 24 (0), or 48 hr (0) before preparationof membranes, and the production of CAMP was measured at theconcentrations of propranolol indicated on the abscissa. The estimatedmembrane concentrations of j32AR in this experiment were 0.04, 1.9,and 10.0 pmol/mg ofprotein at 12, 24, and 48 hr ofinfection, respectively.The data shown are representative of two experiments, and a compa-rable pattern was observed with alprenolol.
a:E
a:a)0
0.
E
a-
0
0
E0.
log [ligand]
Fig. 6. Inhibition of agonist-independent adenylyl cyclase activity inmembranes from Sf9 cells expressing fi2AR. Production of cAMP wasmeasured in the presence of dichloroisoproterenol (#{149}),labetalol (0),pindolol (S), or timolol (0), at the concentrations indicated on the ab-scissa. The representative data shown were obtained on the same daywith the same membrane preparation and were included in the simulta-neous analyseS for the determination of the ECse and E� values listed inTable 1 . Isoproterenol (0) was included as a positive control. For thecurves shown, the data were reanalyzed with y� common to all fivesets of data.
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
494 Chidiac et a!.
TABLE 1Inhibition of fi3AR-associated increase in membrane adenylyl cyclase activity by AR ligandssf9 cells were infected for 48 hr with baculovirus encoding the �32AR, to yield 23 ± 2 pmol of receptor/mg of membrane protein.
flARligands EC�O” n� K/ n E�” n
flu flU
Alprenolol 6.7 ± 1 .1 6 0.32 ± 0.06 3 0.68 ± 0.02 6Propranolol 7.9 ± 1 .2 4 0.36 ± 0.08 3 0.82 ± 0.04 4Timolol 10.0 ± 1.8 6 0.29 ± 0.07 2 0.91 ± 0.04 6Pindolol 36 ± 8 6 0.71 ± 0.01 3 0.65 ± 0.04 6Labetalol 61 ± 44 4 10.6 ± 2.8 2 0.40 ± 0.03 4Dichloroisoproterenol 760 ± 396 4 220 ± 92 1 0.09 ± 0.1 1 4
aARh�fldS KI fi E� fi
1zU
Prazosin e 2.3 ± 1.1 2 �0.7 3Phentolamine e 31 ± 16 2 �0.5 3Yohimbine e 592 ± 83 2 �0.5 3
a Errors accompanying EC� and /(� ValUeS reflect the range over which the sum of squares is insensitive to the value of the parameter.a � number of experiments.
C Values are mean ± standard error.6 SpecifIc binding was not completely inhibited at the highest concentrations of these ligands used (2.0 mM) and nonspecific binding was set equal to that estimated
from concurrent experiments with $AR antagonists.a Maximal effect was not defined at the highest concentrations tested (0.1-1 .0 mM), and therefore EC� could not be estimated reliably.
membranes from cells expressing $2AR was increased 4.5-foldon average, relative to that in membranes from cells lackingthe receptor; none of the agents tested decreased cAMP pro-
duction to a level below that observed in membranes from cells
lacking f32AR. The I3AR ligands tested varied considerably with
respect to their effects on adenylyl cyclase activity, as illus-
trated in Fig. 6. Timolol caused the greatest inhibition and
dichloroisoproterenol the least. Isoproterenol typically stimu-
lated enzymic activity 1.5-2.5-fold.
The inhibitory effects of nine different AR ligands on thebinding of 1251-CYP and on the production of cAMP in mem-
branes from �32AR-expressing Sf� cells are summarized in Table
1. In general, the rank orders of affinity (Kd) and potency(EC�) are in agreement with each other. For each inhibitory
I3AR ligand tested, the ECse value exceeded the K,, value(Table 1).
There was no obvious relationship between the inverse effi-
cacy of a given ligand and its EC�, Kd, or ECse/Kd value. If the
observed inhibition were due to competition between the li-
gands tested and contaminating catecholamines, a correlation
between E1�� and K,, would be expected. Timolol, propranolol,and alprenolol all yielded similar values with respect to potency
and binding affinity, but the inverse efficacy of alprenolol was
significantly less than that of either timolol (p = 0.002) orpropranolol (p = 0.032). Pindolol was less potent than any of
those three agents but exhibited an inverse efficacy similar to
that of alprenolol. Of the I3AR ligands tested, dichloroisopro-terenol was the weakest with respect to potency, affinity, and
inverse efficacy.
Three csAR antagonists, i.e., phentolamine, prazosin, and
yohimbine, were also tested. Although they bound with affini-
ties several orders of magnitude weaker than those of most of
the I3AR antagonists tested, these agents inhibited fl2AR-in-
duced cAMP production by at least 50%, further demonstrating
the lack of corrlation between E1�� and Kd. For example, theinverse efficacy of yohimbine was greater than that of labetalol,even though the Kd of the former was approximately 4 orders
of magnitude higher than that of the latter. Precise values for
ECse and � could not be reliably determined for the aARantagonists, because enzymic activity was still decreasing at
the highest concentrations of those ligands tested (�100 zM)
and � was therefore undefined. Although an analogous prob-
lem was encountered in the binding experiments, values of Kd
could be estimated by assuming that the nonspecific binding of
‘25ICYP was the same with both I3AR and aAR ligands. Com-parison of values of Kd with the incomplete dose-responsecurves obtained with prazosin, phentolamine, and yohimbine
(data not shown) suggests that the rank orders of affinity and
potency are the same for aAR antagonists, consistent with the
effects of I3AR ligands.
Competitive effects among uganda with different in-
verse efficacies. The agonist-independent inhibition of ade-
nylyl cyclase activity by timolol could be partly blocked by less
efficacious agents. At 100 nM timolol, the inhibition of cAMPproduction was nearly maximal (see Fig. 6), but the additional
inclusion oflabetalol (100 �M) partially reversed that inhibition
(Fig. 7A). Similarly, higher concentrations of timolol were
required to inhibit cAMP production in the presence of di-
chloroisoproterenol (100 iM) (Fig. 7B). It is clear that dichlo-
roisoproterenol was not acting as a partial agonist, because itreduced the increase in enzymic activity attributable to the
spontaneous activity of the receptor. The inhibition of theeffect of timolol by ligands with relatively low inverse efficacies
suggests that the different antagonists compete for a common
binding site and that the incomplete effects of ligands such as
labetalol do not arise from a failure to saturate the receptor.The latter point is supported by the observation that both
labetalol and dichloroisoproterenol fully inhibit the specific
binding of ‘9-CYP (Table 1).
Production of cAMP in whole cells. The expression of
132AR was found to be associated with a 2.8-fold increase in
intracellular cAMP in Sf9 cells cultured in serum-free medium(Fig. 8), and a similar increase was observed in cells cultured
in complete medium (data not shown), which is consistent withobservations in washed membranes. This suggests that theincrease in adenylyl cyclase activity observed in membranesreflects the situation in whole cells. Moreover, the agonist-independent increase in cAMP in the absence of serum in Sf9cells expressing the /32AR was inhibited by the I3AR antagonists
propranolol and timolol. In contrast to their inhibitory effects
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
Inverse Agonism at $�AR 495
50
40
30 #{176}-40
ELC) a-
0
10 _5E0.
0 0
TimololLabetalol log [Timolol]
Fig. 7. Effect of agents with low E� on inhibition of adenylyl cyclase activity by timolol. A, Proportion of adenylyl cyclase activity associated with$2AR expression observed in the presence of 10 n� timolol, 1 0 MM labetalol, or both. The results shown represent means from two experiments. B,Effect of timolol on the production of cAMP in membranes from Sf9 cells expressing fl2AR, in the absence (0) and presence (0) of 10 �Mdichloroisoproterenol. The data shown are representative of two experiments.
I ,2
1,0
+ +
- + +
-10 -8 -6 -4
0,8
0,6
0,4
0,2
0,0
#{149}0,________� v),�TT____
I I
1 2 3 4 5 6 7 8 9 10
Fig. 8. Effect of $AR ligands on intracellular levels of CAMP in Sf9 cells expressing fl2AR. Cells were treated for 30 mm before assay of intracellularCAMP concentrations. The data from each experiment were scaled by dividing the level of intracellular CAMP observed under each condition by thatobserved in the presence of 1 �M isoproterenol in the same experiment (the average of the latter from four experiments was 17.1 ± 6.6 pmol/sample), and the data shown represent the means of two to four independent observations. Error bars, standard deviations. 1-9, Sf9 cells infectedwith recombinant baculovirus encoding the �32AR; 10, Sf9 cells infected with wild-type baculovirus. Treatments were as follows: 1, 1 �M isoproterenol(four experiments); 2, 10 �iM dichloroisoproterenol (three experiments); 3, 1 0 �M labetalol (two experiments); 4, 1 0 �M pindOIOI (three experiments);5, no treatment (four experiments); 6, 10 �M aiprenolol (four experiments); 7, 1 �M isoproterenol plus I 0 MM aiprenolol (four experiments); 8, 1 0 �M
propranolol (four experiments); 9, 10 �M timolol (three experiments); 10, no treatment (three experiments).
a:
0
0
�00
0.
a-
0
�0a)0
EU)
a.
4U
.
00
on membranes, aiprenolol and pindolol had minimal effects on (Fig. 9). It follows that the phenomena observed in whole cellsintracellular cAMP. Alprenolol was slightly inhibitory, and and in membranes may reflect the same underlying mecha-
pindolol was slightly stimulatory. Furthermore, both labetalol
and dichloroisoproterenol, which were relatively poor inhibitors
in membranes, were clearly stimulatory in whole cells.
Surprisingly, the present results suggest that the same ligand
can stimulate the activity of a receptor under one set of exper-
imental conditions and inhibit the same receptor under a
different set of conditions. Still, the ligands tested revealed the
same rank order with respect to their effects in whole cells and
in membranes. Moreover, although values of E1�� in whole cells
were found to be negative for three of the six ligands (stimu-
latory ligands by definition have negative E1�� values), a signif-
icant correlation was found with E1�� in membranes ( r� = 0.94)
nisms.
Effect of forskolin on cAMP production and inverse
agonism. The apparent loss of inverse agonism in whole cells
with some ofthe ligands tested suggests that the “basal” activity
of adenylyl cyclase may have differed between the two prepa-
rations tested. It follows that inverse agonism may be depend-
ent upon not only the activation state of the receptor but also
that of the effector. Experiments therefore were carried out to
determine whether activation of adenylyl cyclase by forskolinwould alter the $AR responses observed in Sf9 cells expressingthe f�2AR. Adenylyl cyclase activity was stimulated by forskolin
(1 �M) in Sf9 cells expressing �32AR and in membranes derived
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
0
0
U
.0
USU
.
0
0
C
0,0 0,2 0,4 0,6 0,8 1,0
Inverse efficacy (membranes)
B>�
>.4�I
0
0
a)U)
00>‘
0
>‘
>‘a:
�00
a)>
0
0.6
0.3
0.0
a)>a)
a-
0
a)>
0
a)
0.4
0.2
0.0
Fig. 10. Effect of $AR ligands on fl2AR activity in the presence of forskolin. Membrane cAMP production (A) and intracellular cAMP concentration(B) were measured in the presence of 1 �iM forskolin in Sf9 cells infected for 48 hr with either wild-type (0) or recombinant baculovirus encoding the
I92AR in the absence of I9AR ligands (U), in the presence of 10 �iM propranolol (), or in the presence of 10 �M pindolol (U). Data from individualexperiments were scaled as a fraction of the signal observed in the presence of 1 �M isoproterenol, as described in the legend to Fig. 8 (membranes,70 ± 40 pmol/mg of protein/mm; whole cells, 8.8 ± 2.5 pmol of cAMP/sample). The scaled data from two independent experiments in A and threeindependent experiments in B were averaged to yield the values indicated. Error bars, standard deviations. Additional details are described inMaterials and Methods.
496 Chidiac et al.
Fig. 9. Correlation between values of � obtained in membranes (Table1) and whole cells (Fig. 8) for the following ligands: 1 , dichloroisoproter-enol; 2, labetalol; 3, pindolol; 4, alprenolol; 5, propranolol; 6, timolol.
from those cells. Analogously to observations in the absence of
forskolin, expression of the f32AR was associated with increases
in both intracellular cAMP and membrane adenylyl cyclaseactivity in the presence of forskolin (Fig. 10). Similarly, the
effects of pindolol and propranolol were comparable in assays
carried out in the presence and absence of forskolin; in both
cases, pindolol was inhibitory in membranes (Figs. 6 and 1OA)
and slightly stimulatory in whole cells (Figs. 8 and lOB),
whereas propranolol was inhibitory in membranes (Figs. 5 and
lOA) as well as in whole cells (Figs. 8 and lOB). Forskolin thus
appeared to have little influence on the overall effects of either
the unliganded f32AR or the inverse agonists tested on adenylyl
cyclase activity in Sf9 cells.Inhibition of receptor-stimulated cAMP production in
mammalian cells. To ascertain whether the inverse agonist
effects observed in Sf9 cells are also characteristic of f32AR
expressed in mammalian systems, experiments were carried outusing membranes from CHW cells expressing human fl2AR.
Agonist-independent adenylyl cyclase activity was inhibited in
a dose-dependent manner by labetalol, pindolol, alprenolol,
propranolol, and timolol (Fig. 11). The observed rank order of
these ligands with respect to potency in CHW cells is the same
as that obtained using the baculovirus/Sf9 expression system,
as is the rank order with respect to inverse efficacy (Table 1).
Discussion
The results of the present study show that the agonist-
independent, spontaneous activity of a G protein-coupledreceptor in intact cells can be inhibited by antagonists (Fig. 8),
suggesting that the pharmacological effects of the latter mayarise at least partly from their direct interaction with receptors.
The flAR ligands tested inhibited adenylyl cyclase in mem-branes to varying extents, as indicated by the range of E1��values observed. As discussed below, it appears that the E1�� of
a drug determined in membranes may reflect its pharmacolog-ical properties in vivo.
One potential caveat in studying the intrinsic effects of
antagonists is that contaminating catecholamines may have
been present in the serum included in the cell culture medium
(19), but several observations indicate that both the �92AR-
induced increase in adenylyl cyclase activity and the inhibition
of that activity by antagonists are independent of the possible
influence of such contaminants. 1) It is estimated that the
soluble constituents in the culture medium were diluted at least
107-fold during the routine preparation of the washed mem-
branes. 2) Both the increase in adenylyl cyclase activity (Fig.
1, inset, and Fig. 8) and the inhibition of the enzyme byantagonists were observed in the absence of serum (Figs. 4 and
8). 3) Alprenolol inhibited the enzyme in membranes preparedfrom cells that had been cultured in the absence of serum and!
or exposed to alprenolol before harvesting (Fig. 4). 4) E1�� was
not correlated with binding affinity (Table 1), a relationship
that would be expected if residual catecholamines were respon-
sible for the increase in adenylyl cyclase activity. 5) The effect
of timolol was inhibited both by labetalol and by dichioroiso-
proterenol (Fig. 7); if inhibition by timolol were due to the
displacement of bound activating ligand, an additional inhibi-tory ligand would not be expected to decrease that effect.
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
-4...a:
E
a:a)
0
0.
E
a-
0
0
E0.
15
12
9
6
3
0
-11 -9 -7 -5
log [ligand]
Inverse Agonism at $2AR 497
Fig. I 1. Inhibition of agonist-independent adenylyl cyclase activity inmembranes from CHW cells expressing /32AR. Production of cAMP wasmeasured in the presence of labetalol (#{149}),pindolol (0), alprenolol 4,propranolol (0), or timolol (#{149}),at the concentrations indicated on theabscissa. The data presented represent the mean values of three exper-iments carried out in duplicate. The averaged data sets were analyzedsimultaneously, assuming a common value of Yx-o The fitted values ofEC� are as follows: labetalol, 1 .1 ± 1 .7 z�i; pindolol, 13 ± 10 nM;alprenolol, 1 .3 ± 0.7 nM; propranolol, 2.0 ± 0.8 nM; timolol, 2.5 ± 0.9nM.
It seems likely that the interactions between /.3AR ligands
and the fl2AR revealed using the baculovirus/Sf9 expression
system accurately mimic the process of I.�AR transduction inmammalian systems. I3AR expressed in Sf9 cells exhibit normalpharmacological properties and undergo phosphorylation and
palmitoylation in the expected manner (11, 20). Furthermore,
the binding affinities of antagonists obtained in the present
study are consistent with those reported in membranes pre-
pared from flAR-expressing mammalian cells (10, 21) and tis-
sues (22-24). Inverse agonism has been demonstrated in this
study in both CHW and Sf9 cells, indicating that the phenom-
enon can occur in both mammalian and insect cells and is not
dependent upon the baculovirus/Sf9 expression system per se.
The rank order of I3AR ligands, moreover, appears to be the
same with respect to both potency and inverse efficacy in both
systems (Fig. 11; Table 1).
Spontaneous activity of $2AR in Sf9 cells. The results
shown in Figs. 1 and 2 indicate that the $2AR activates adenylyl
cyclase in an agonist-independent manner in membranes from
Sf9 cells. In accord with the present results, purified I3AR was
shown to stimulate the GTPase activity of G, in a reconstituted
system, even in the absence of agonist (25). Also, the effect of
the �32AR on intracellular cAMP in Sf9 cells (Fig. 8) agrees
with the results of a previous study (10) in showing that wild-
type f32AR can exhibit spontaneous activity in intact systems.These findings indicate that I�AR can assume an activated stateeven when the ligand binding site is vacant.
Spontaneous activity also has been observed in membranes
from cells expressing #{244}-opioid receptors, as implied by the
inhibitory effects of some antagonists on basal GTPase activity(4). In contrast, the same antagonists were found to be without
effect on the production of cAMP in intact cells in the absence
of agonist (5). It has been suggested that the reversal of spon-
taneous receptor activity in an intact system by antagonists
would support the idea that the biological effects of such ligands
reflect their negative intrinsic activity (2). The present results
show that spontaneous activity of an unliganded G protein-
coupled receptor can occur in whole cells and that this activity
can be inhibited by antagonists (Fig. 8). This suggests that thephysiological effects of some antagonists can be ascribed at
least partially to their negative intrinsic activity.Negative intrinsic activity of I�AR ligands. The results
of the present study reveal that different flAR antagonists
inhibit to varying degrees the spontaneous activity of the (32AR
in membranes (Fig. 6; Table 1). These ligands generally are not
reported to inhibit basal adenylyl cyclase activity in such prep-
arations (e.g., Refs. 26-28). However, in accord with the present
results, propranolol, timolol, and pindolol were found to inhibit
guanosine-5’ -(fl,-y-imido)triphosphate-stimulated adenylyl cy-
clase in membranes from turkey erythrocytes (7). In contrast
to the present results, dichloroisoproterenol has been found in
some cases to stimulate cAMP production in membranes
(28, 29).
Antagonists that yielded the highest values of E1�� in mem-
branes (i.e., timolol and propranolol) decreased intracellular
cAMP, whereas those exhibiting the lowest values of E1�� in
membranes (i.e., dichloroisoproterenol and labetalol) wereclearly stimulatory in whole cells (Fig. 8). flAR antagonists
generally are found to be without effect on intracellular cAMP
(10, 21, 30-32), although dichloroisoproterenol, pindolol, and
aiprenolol have been reported to stimulate adenylyl cyclase
activity in intact cells in some studies (10, 21). In contrast tothe present results, previous studies seemingly have failed to
detect the inhibition of spontaneous fl2AR activity in whole
cells.In both membranes and whole cells, the inconsistent obser-
vations with flAR antagonists may reflect the difficulty of
measuring the effects of these ligands on spontaneous receptor
activity. The feasibility of quantifying such effects constitutes
one advantage of the baculovirus system, owing to the relatively
high spontaneous activity associated with high levels of /32AR
expression. If “neutral” results are omitted, the available data,
including the present results, may be summarized as follows:
1) propranolol and timolol inhibit adenylyl cyclase both in
membranes and in whole cells, 2) alprenolol, labetalol, and
pindolol inhibit activity in membranes and stimulate activity
in whole cells, and 3) dichloroisoproterenol is inhibitory or
stimulatory in membranes and stimulatory in whole cells.
In membranes from Sf9 cells, the net stimulation of adenylyl
cyclase by agonists and the net inhibition of that activity by
inverse agonists both are dependent upon the density of �32AR
(Figs. 2 and 5, respectively). The present results suggest that
more receptors may be required for the detection of inhibition
than are needed to observe stimulation, because a slight stim-
ulatory response to isoproterenol usually could be detected 12
hr after infection of cells with the recombinant baculovirus(approximately 30 fmol of �32AR/mg of membrane protein),
whereas inverse agonism was first evident after a 24-hr infec-tion (1-2 pmol of f32AR/mg of membrane protein). This appar-
ent discrepancy is not surprising, because inhibition is inher-
ently more difficult to detect than stimulation. It is likely thatinverse agonism occurs at low receptor levels, even though it is
difficult to quantitate. Alternatively, one might interpret the
apparent requirement for large numbers of receptors as evi-
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
498 Chidlac et aL
1 p� Chidiac, S. Parent, and M. Bouvier, unpublished observations.2 p� chidiac and J. W. Wells, unpublished observations.
dence that inverse agonism does not occur in nature, because
endogenous levels of G protein-coupled receptors would be
below the required threshold. Several lines of evidence, how-ever, argue against that notion. 1) Inverse agonism has been
observed in preliminary experiments in this laboratory with
two other mammalian cell lines, expressing approximately 200and 800 fmol of fl2AR/mg of protein.’ 2) Endogenous receptorsappear to be regulated directly by antagonists, because inverse
agonism in membrane preparations has been observed with
$AR in avian erythrocytes (7) and muscarinic receptors in
porcine (9) and hamster hearts.2 3) The specific binding of 125I�
lysergic acid diethylamide to serotonergic sites in rat choroid
plexus homogenates reveals a receptor density of 3.1 pmol/mg
of protein (33), which is consistent with the density of fl2AR
associated with inverse agonism in the present study. Moreover,
although G protein-coupled receptors typically are found tooccur at femtomole/milligram of protein levels in homogenates
of mammalian tissue, the actual densities in individual neurons,
or indeed at synaptic clefts, are uncertain. 4) Levitzki (34) hascalculated that agonist occupancy of <0.02% of flAR in turkey
erythrocyte or rat parotid gland is sufficient to saturate the
relevant protein kinase and protein kinase-dependent processes
that lead to agonist-induced changes in cellular activity; itfollows that a level of spontaneous receptor activity too low to
quantitate reliably may still have a substantial impact oncellular function. Taken together, the foregoing points attestto the potential importance of the inhibition of spontaneous
receptor activity by inverse agonists in the pharmacological
regulation of G protein-coupled receptors in vivo.
Several of the f�AR ligands used in the present study arereported to act as partial agonists in vivo. Alprenolol, dichlo-
roisoproterenol, labetalol, and pindolol all exhibit ISA (21, 35),a phenomenon wherein I3AR blockade is observed at high rates
of sympathetic activity but stimulatory effects, such as in-
creased heart rate, may be observed at rest (21, 36). Timololand propranolol, which lack ISA, are inhibitory in both mem-
branes and whole cells. In contrast, aiprenolol, pindolol, labe-talol, and dichloroisoproterenol, ligands with ISA, tend to be
inhibitory in membranes (7) (Figs. 3-7 in this study) andstimulatory in whole cells (10, 21) (Fig. 8 in this study). In light
of such observations, it is unclear whether these drugs are
agonists or antagonists; moreover, the difference between anagonist and an antagonist no longer seems obvious. It has been
suggested that the efficacy of a ligand at a particular receptor
can vary depending upon which G protein is coupled to the
receptor (5), and this may account for some of the observeddifferences between cell types. However, it is unlikely that it
could explain the differences between membranes and wholecells in the present study, because the same cells were used forboth types of experiment.
The correlation observed between values of � in mem-branes and whole cells (Fig. 9) may be of some help in resolving
the present paradox. Although it is not understood how one
ligand is able to stimulate the activity of a receptor under one
set of experimental conditions and inhibit the same receptor
under a different set of conditions, the correlation suggests that
similar mechanisms may underlie both processes. The increasein intracellular cAMP in Sf9 cells occurred only with ligands
exhibiting ISA, whereas those lacking ISA were inhibitory. Inaddition, E1�� in membranes is inversely correlated with ISA
(r2 = -O.94).� In practical terms, the effects of a novel $AR
ligand on adenylyl cyclase activity in fl2AR-expressing Sf9 cells
and in membranes derived from those cells may be of value in
predicting its level of ISA in vivo. There was no obvious
relationship between E1�� and either the lipophilicity or the
local anesthetic (“membrane-stabilizing”) properties of the li-
gands tested (37).
Mechanistic implications. The two-state scheme shown
below depicts a system wherein a receptor spontaneously as-
sumes either an inactive state (R) or an active state (R’), withthe equilibrium between these being described by the unimo-
lecular constant K� ([R’J/[R]).
K,.
L+R � LR
1�Ks lLaKsaKL
L + R* � LR*
A given ligand (L) binds to R and R, to yield the association
constants KL (i.e., [LR]/[L][R]) and aKL (i.e., [LR’J/[LJ[R’J),respectively; L is assumed to have no other effect on either R
or R’. If a is >1, then L binds with greater affinity to the activeform of the receptor and increases the number of receptors inthe active state (i.e., [R’] + [LW]), thereby stimulating the
activity of the receptor (i.e., L is an agonist). If a is <1, the
number of receptors in the inactive state is increased by L (i.e.,
L has negative intrinsic activity). The influence of any auxil-liary proteins is not directly considered. This model is analo-
gous to that proposed for the benzodiazepine receptor (1, 3, 38)
and also to the ternary scheme proposed for G protein-coupled
receptors (5), at least under conditions where the ligand doesnot appreciably affect the concentration of free G protein.
At the root of such models is the notion that the biochemicaleffect of a ligand-receptor interaction reflects the relative pref-
erence of a ligand for the “active” or the “inactive” form of the
receptor. The stimulatory effects of agonists thus are attributed
to their preference for the active receptor, which leads to anincrease in the proportion of receptors in that form. The effectof antagonists can be either negative or neutral; those that
favor the inactive form of the receptor exhibit negative intrinsicactivity, and those with no preference lack intrinsic activity. In
the presence of an activating ligand, both types of antagonistappear inhibitory. Also, in the absence of any receptor in theactive form, both appear to be without effect.
Not all ligands tested in the present study inhibit cAMP
production in membranes to the same extent (Fig. 6; Table 1),
and their effects in whole cells similarly exhibit a range ofefficacies (Fig. 8). Taken separately, each of these observationsis consistent with the notion that a ligand can modulate the
function of a receptor by binding preferentially to either the
active or the inactive state. Taken together, however, the pres-ent results suggest that some ligands can act as agonists (i.e.,
K,. < aKL) in whole cells and as antagonists (i.e., KL > aKL) inmembranes prepared from the same cells; it is difficult to
reconcile this observation with the notion that a ligand pro-
3 The correlation coefficent was calculated using ISA values listed by Wailer(35) for alprenolol, pindolol, and labetaiol (dilevalol), assuming an ISA of 0.0 forboth propranolol and timolol (35, 39).
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from
Inviris A�onhimit &AR 4�duces its effect by binding preferentially to either the active or
the inactive form of a receptor, particularly if the only effect of
the ligand is to “stabilize” one of two states.
Because expression of the f.�2AR is associated with a 4.5-fold
increase in adenylyl cyclase activity in membranes but only a
2.8-fold increase in whole cells, it appears that the unliganded
receptor may be more “active” in membranes than in whole
cells. In terms of the model described above, K� is greater in
membranes than in whole cells. An analogous conclusion was
reached with respect to the function of the t5-opioid receptor inmembranes and whole cells when inverse agonism was not
reliably observed in the latter (4, 5). Such a difference between
membranes and whole cells could account for the apparent lossof activity of an antagonist in cells, as was observed with
alprenolol in the present study. The divergent effects of labe-talol and dichloroisoproterenol in membranes versus whole
cells, however, may be indicative of a more complex system. In
terms of the scheme shown above, a single ligand will inhibit a
receptor under one set of conditions and stimulate it under
another set of conditions only if the ligand binds preferentially
to R’ in one case and to R in the other. Alternatively, more
than two states of the f32AR may be involved in mediating the
effects of f.�AR ligands.Although any configuration of the fl2AR occurring in mem-
branes presumably could arise in whole cells, the converse isnot necessarily true. The discrepancies observed between the
effects of I3AR ligands in whole cells and membranes thus may
reflect the existence of some receptor-associated component(s)
present in the former but absent from the latter. It is possiblethat, in addition to the soluble constituents of the cell, somemembrane-associated components were lost during the prepa-
ration of the membranes used in the present study. Current
studies in this laboratory are aimed at identifying any cellular
constituents that may, when added to membrane adenylyl
cyclase assays, lead to I3AR responses that more accurately
reflect signal transduction in whole cells.
Acknowledgments
The authors thank Drs. J. W. Wells and A. Fargin for helpful discussions.
References
1. Ehlert, F. J., W. R. Roeske, K. W. Gee, and H. I. Yamamura. An allostericmodel for benzodiazepine receptor function. Biochem. Phormo.coL 32:2375-2383 (1983).
2. Sch#{252}tz,W., and M. Freissmuth. Reverse intrinsic activity of antagonists onG protein-coupled receptors. Trends Pharmacol. Sci. 13:376-380 (1992).
3. Ehlert, F. J. ‘Inverse agonists,’ cooperativity and drug action at benzodiaze-pine receptors. Trends Pharrnacol. Sci. 7:28-32 (1986).
4. Costa, T., and A. Herz. Antagonists with negative intrinsic activity at #{244}-opioid receptors coupled to GTP-binding proteins. Proc. Nati. Aco.d. Sci. USA
86:7321-7325 (1989).5. Costa, T., Y. Ogino, P. J. Munson, H. 0. Onaran, and D. Rodbard. Drug
efficacy at guanine nucleotide-binding regulatory protein-linked receptors:thermodynamic interpretation of negative antagonism and of receptor affin-ity in the absence of ligand. Mol. PharmacoL 41:549-560 (1992).
6. Braestrup, C., R. Schmiechen, G. Neef, M. Nielsen, and E. N. Petersen.Interaction of convulsive ligands with benzodiazepine receptors. Science(Washington D. C.) 216:1241-1243 (1982).
7. Murray, R., and A. K. Keenan. The f�-adrenoceptor is precoupled to G� inchicken erythrocyte membranes. Cell. Signalling 1:173-179 (1989).
8. Freissmuth, M., E. Selzer, S. Marullo, W. Sch#{252}tz,and A. D. Strosberg.Expression of two human fl-adrenergic receptors in Escherichw coli: func-tional interaction with two forms of the stimulatory G protein. Proc. Nati.Acad. Sci. USA 88:8548-8352 (1991).
9. Hilf, G., and K. H. Jakobs. Agonist-independent inhibition of G proteinactivation by muscarinic acetylcholine receptor antagonists in cardiac mem-branes. Eur. J. Pharrnacol. 225:245-252 (1992).
10. Samama, P., S. Cotecchia, T. Costa, and R. J. Lefkowitz. A mutation-inducedactivated state of the �92-athenergic receptor. J. BiOI. Chem. 268:4625-4636(1993).
11. Mouillac, B., M. Caron, H. Bonin, M. Dennis, and M. Bouvier. Agonist-
modulated palmitoylation of�i,-athenergic receptor in Sf9 cells. J. BIOI. Chem.267:21733-21737 (1992).
12. Bradford, M. M. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye binding.Anal. Biochem. 72:248-254 (1976).
13. Moffett, S., B. Mouillac, H. Bonin, and M. Bouvier. Altered phosphorylationand desensitization patterns of a human �3,-adrenergic receptor lacking thepalmitoylated Cys-341. EMBO J. 12:349-356 (1993).
14. Salomon, Y., C. Londos, and M. Rodbell. A highly sensitive adenylate cyclaseassay. Anal. Biochem. 58:541-548 (1974).
15. Gilman, A. G. A protein binding assay for adenosine 3’ :5’ cyclic monophos-phate. Proc. NatL Aced. Sci. USA 67:305-312 (1970).
16. mae!, P. A., and M. Sanda. Temperature-dependent changes in binding to �f-adrenergic receptors of intact S49 lymphoma cells. J. BiOL. Chem. 254:6554-6559 (1979).
17. De Lean, A., P. J. Munson, and D. Rodbard. Simultaneous analysis of familiesof sigmoidal curves: application to bioassay, radioligand assay, and physio-logical dose-response curves. Am. J. Physiol. 235:E97-E102 (1978).
18. De Lean, A., A. A. Hancock, and R. J. Lefkowitz. Validation and statisticalanalysis of a computer modeling method for quantitative analysis of radioli-gand binding data for mixtures of pharmacological receptor subtypes. MoLPharrnocoL 21:5-16 (1982).
19. Dibner, M. D., and P. A. Insel. Growth of C6 glioma cells in serum-containingmedium decreases �3-adrenergic receptor number. J. Cell. Physiol. 109:309-315 (1981).
20. Parker, B. M., K. Kameyama, T. Higashijima, and E. M. Ross. Reconstitu-tively active G protein-coupled receptors purified from baculovirus-infectedinsect cells. J. BiOL Chem. 266:519-527 (1991).
21. Jasper, J. R., M. C. Michel, and P. A. Insel. Molecular mechanism of �f-adrenergic receptor blockers with intrinsic sympathomimetic activity. FA -
SEB J. 2:2891-2894 (1988).22. Bevilacqua, M., A. Monopoli, T. Vago, G. Baldi, E. Ongini, and G. Norbiato.
Characterization of dilevalol affinity for fi- and a-adrenergic receptors ofhuman cardiovascular tissues. J. Cardiovasc. PharmacoL 1 1 (Suppi. 2):S25-S27 (1988).
23. Riva, E., T. Mennini, and R. Latini. The a- and �9-adrenoceptor blockingactivities of labetalol and its RR-SR (50:50) stereoisomers. Br. J. Pharmo.coL104:823-828 (1991).
24. Minneman, K. P., L. R. Hegstrand, and P. B. Molinoff. The pharmacologicalspecificity of beta-i and beta-2 adrenergic receptors in rat heart and lung invitro. MoL PharrnacoL 16:21-33 (1979).
25. Cerione, R. A., D. R. Sibley, J. L. Benovic, M. G. Caron, and R. J. Leflwwitz.Reconstitution of a hormone-sensitive adenylyl cyclase system: the pure �-
adrenergic receptor and guanine nucleotide regulatory protein confer hor-mone responsiveness on the resolved catalytic unit. J. BiOL Chem. 259:9979-9982 (1984).
26. Golf, S., and V. Hansson. Relative potencies of various �3-adrenoceptorantagonists (BAA) at the level of the human myocardial f3-adrenoceptor-adenylate cyclase (AC) complex. Is intrinsic sympathomimetic activity (ISA)due to AC activation? Scand. J. Clin. Lab. Invest. 46:121-130 (1986).
27. Reynolds, E. E., and P. B. Molinoff. Down-regulation of fl-adrenergic recep-tors in 549 lymphoma cells induced by atypical agonists. J. PharrnacoL Exp.Ther. 239:654-660 (1986).
28. Strader, C. D., M. R. Candelore, W. S. Hill, R. A. F. Dixon, and I. S. Sigal.A single amino acid substitution in the fl-adrenergic receptor promotes partialagonist activity from antagonists. J. BiD!. Chem. 264:16470-16477 (1989).
29. Benovic, J. L., C. Staniszewski, F. Mayor, Jr., M. G. Caron, and R. J.Lefkowitz. fl-Adrenergic receptor kinase: activity of partial agonists forstimulation of adenylate cyclase correlates with ability to promote receptorphosphorylation. J. BaiL Chem. 263:3893-3897 (1988).
30. Hedberg, A., E. Carlsson, B. Fellenius, and B. Lundgren. Cardiostimulatoryeffects of prenalterol, a beta-i partial agonist, in vivo and in vitro. Naunyn-Schmtedebergs Arch. PharmacoL 3 18:185-191 (1982).
31. Neve, K. A., D. A. Barret, and P. B. Molinoff. Selective regulation of fly-and
fl-adrenergic receptors by atypical agonists. J. PharmacoL Exp. Ther.235:657-664 (1985).
32. Michel, M. C., X. L. Wang, A. Daul, and O.-E. Brodde. fl-Adrenoceptorantagonists with intrinsic sympathomimetic activity down-regulate humanlymphocyte fl-adrenoceptor density but do not activate the lymphocyte cyclicAMP system. J. Cardiovasc. Pharmo.coL 10 (Suppl. 4):S117-S119 (1987).
33. Yagaloff, K. A., and P. R. Hartig. 1’�I-Lysergic acid diethylamide binds to anovel serotonergic site on rat choroid plexus epithelial cells. J. Neurosci.5:3178-3183 (1985).
34. Levitzki, A. Catecholamine receptors, in Cellular Receptors for Hormones andNeurotransmitters (D. Schuster and A. Levitzki, eds.). John Wiley & Sons,New York, 267-286 (1980).
35. Wailer, D. G. �-Adrenoceptor partial agonists: a renaissance in cardiovasculartherapy? Br. J. Clin. PharmacoL 30:157-171 (1990).
36. Dollery, C. T. fl-Adrenoceptor blockade: past, present, and future. J. Cardio-vasc. PharrnacoL 1 1 (Suppl. 2):S1-S4 (1988).
37. Flattery, K. V., and L. Spero. Adrenoceptor antagonists, in Principles ofMedical Pharmacology (H. Kalant, W. H. E. Roschlau, and E. M. Sellers,eds.). Oxford University Press, New York, 195-202 (1985).
38. Chiu, T. H., D. M. Dryden, and H. C. Rosenberg. Kinetics of [3Hjflunitra-zepam binding to membrane-bound benzodiazepine receptors. Mol. Phar-rnacoL 21:57-65 (1982).
39. McDevitt, D. G. Pharmacological characteristics of �i blockers and their rolein clinical practice. J. Cardiovasc. PharmacoL 8 (Suppl. 6):S5-S11 (1986).
Send reprint requests to: Michel Bouvier, D#{233}partement de biochimie, Univ.ersit#{233}de Montr#{233}al, P.O. Box 6128, Station A, Montr#{233}al, Qu#{233}bec,H3C 837 Canada.
at ASPE
T Journals on A
ugust 30, 2016m
olpharm.aspetjournals.org
Dow
nloaded from